| Contents |
| In the talk, the ill-posed problem of solving linear equations in the
space of vector-valued finite Radon measures and Hilbert-space data
is addressed. Well-posedness of Tikhonov-regularization with the Radon norm
as well as further regularization properties and optimality conditions
are discussed. Moreover, a flexible and convergent optimization algorithm
in the space of measures is proposed.
As an example, analysis and numerical experiments for sparse deconvolution
are presented. For this problem, optimization in the space of Radon measures
turns out to be a suitable and effective approach. |
|